Subject: Re: Another changer, another changer problem
To: Brett Lymn <blymn@baea.com.au>
From: Bill Studenmund <skippy@macro.stanford.edu>
List: current-users
Date: 10/07/1998 11:03:14
On Wed, 7 Oct 1998, Brett Lymn wrote:

> According to David Maxwell:
> >
> >Currently, the requirement to build a kernel for each change makes the
> >whole wiring down suggestion echo around here a bit troublesome.
> 
> Ummm you don't _have_ to just wire what you have.  You can build in
> support for some devices that you may reasonably expect to add.  On
> the SunOS machines I run you can reasonably expect that scsi id's 4 &
> 5 are tapes and 6 is the cdrom - regardless of whether they are there
> or not (these are the "normal" id's for these devices on a Sun).  I do
> the same on my NetBSD machines.  Alternatively, you can wire down what
> you _do_ have and leave a catch-all auto-probe line in to support an
> emergency device add, this means that you can keep your working
> devices where they are and add other devices on the fly.

And don't forget you can also do:

cd0 at scsibus0 target 5 lun 0
sd5 at scsibus0 target 5 lun 0
st1 at scsibus0 target 5 lun 0

Only one of cd0,sd5,st1 will attach. So if you have a cd in some machines
and a tape in others, you can use the same kernel for both.

:-)

Take care,

Bill